Arabidopsis heat shock transcription factor HSFA7b positively mediates salt stress tolerance by binding to an E-box-like motif to regulate gene expression.

Plant heat shock transcription factors (HSFs) are involved in heat and other abiotic stress responses. However, their functions in salt tolerance are little known. In this study, we characterized the function of a HSF from Arabidopsis, AtHSFA7b, in salt tolerance. AtHSFA7b is a nuclear protein with transactivation activity. ChIP-seq combined with an RNA-seq assay indicated that AtHSFA7b preferentially binds to a novel cis-acting element, termed the E-box-like motif, to regulate gene expression; it also binds to the heat shock element motif. Under salt conditions, AtHSFA7b regulates its target genes to mediate serial physiological changes, including maintaining cellular ion homeostasis, reducing water loss rate, decreasing reactive oxygen species accumulation, and adjusting osmotic potential, which ultimately leads to improved salt tolerance. Additionally, most cellulose synthase-like (CSL) and cellulose synthase (CESA) family genes were inhibited by AtHSFA7b; some of them were randomly selected for salt tolerance characterization, and they were mainly found to negatively modulate salt tolerance. By contrast, some transcription factors (TFs) were induced by AtHSFA7b; among them, we randomly identified six TFs that positively regulate salt tolerance. Thus, AtHSFA7b serves as a transactivator that positively mediates salinity tolerance mainly through binding to the E-box-like motif to regulate gene expression.

Introduction

Plant heat shock transcription factors (HSFs) positively or negatively regulate transcription, and are commonly involved in responses to heat stress and other abiotic stresses, including salinity, cold, and drought (Swindell ; Hu ). Although the size and sequence of plant HSFs are highly variable, their function and structure are conserved (Lin ). For instance, all HSFs have a core structure that contains oligomerization and DBD domains (DNA binding), which is the most conserved region at the N-terminus and comprises a three α-helix bundle and an antiparallel four-stranded β-sheet (Baniwal ). Another conserved domain is an H2–T–H3 (helix–turn–helix) structure serving as the heat shock element (HSE) motif recognition sequence (Lin ; Liu ; Scharf ). The oligomerization domain has a HR-A/B (hydrophobic coiled-coil region) comprising hydrophobic heptad repeats (Liao ). Additionally, some HSFs contain other domains, including C-terminal activation domains (CTADs), nuclear export signal (NES), and nuclear localization signal domains (NLSs) (Baniwal ). There are some aromatic/hydrophobic/acidic (AHA) motifs in the CTADs rich in acidic, aromatic and hydrophobic amino acids (Kotak ; Liao ). Plant HSF proteins mainly include three subfamilies (class A, B, and C) based on the flexible linkers and HR-A/B regions. There are extended HR-A/B regions in class C and A. There are 7 and 21 amino acid residues respectively between the B and A class of the regions of HR-A/B; however, in class B HSFs, the region of HR-A/B is compact and shorter than those in class C and A. Additionally, there is a heptad repeat pattern in class B HSFs that is distinct from those of class C and A (Nover ; Scharf ; Liao ).

HSF proteins have been identified from many organisms, but are mainly found in plant species, with only a few found in yeast, fruit fly, and vertebrates. There are 15–56 HSF genes identified in different plants. For instance, 17 HSFs are found in the genome of peach (Qiao ), 21 in Arabidopsis, 25 in rice, 30 in maize, 52 in soybean (Scharf ), 33 in European pear (Qiao ), and about 56 in wheat (Xue ).

The expression of HSFs is found to be regulated by heat and other abiotic stresses, such as drought, cold, and salt, indicating their roles in these abiotic stresses. In addition, previous studies showed that plant HSFs are involved in various signaling pathways of abiotic stresses, or play functional roles in development (Hwang ). For instance, Arabidopsis HSFA1 serves as a ‘master regulator’ in the heat shock response (Ohama ). Knockdown or knockout of HSFA1 in tomato and Arabidopsis both induced a heat shock-sensitive phenotype and correspondingly reduced the expression of many heat shock responsive genes (Mishra ; Liu ; Yoshida ). HSFA1 can bind to the promoters of HEAT-INDUCED TAS1 TARGET1 (HTT1) and HTT2 to activate their expression, and both of them are involved in thermotolerance (Li ). In addition, HSFA1 mediates thermotolerance through directly regulating the transcription factors (TFs) involved in the heat shock response, including dehydration responsive element-binding factor 2A (DREB2A), HSFB, HSFA2, and HSFA7a (Yoshida ). When HSFA3 was knocked out or knocked down in Arabidopsis, the transcripts of HSP genes were highly decreased under heat stress conditions, suggesting that HSFA3 is involved in the heat shock stress response (Schramm ; Yoshida ). HSFA2 is essential for the heat shock response in plants and is directly regulated by HSFA1, and hsfa2-knockout mutant plants displayed high sensitivity to heat shock and decreased transcripts of a series of heat shock-inducible genes (Charng ). In Arabidopsis, HSFB2b represses the morning clock gene PSEUDO-RESPONSE REGULATOR 7 (PRR7), which plays a role in salt and heat signaling input to the circadian clock, and is involved in elevating tolerance to stress (Kolmos ).

HSFs are found to play their roles in transcriptional regulation by binding to HSEs (5′-AGAAnnTTCT-3′) (Akerfelt ). An HSF (PeHSF) from Populus euphratica is induced by salinity and can bind to the HSE in the promoter of PeWRKY1 to activate its expression, leading to improved tolerance to salt (Shen ). Additionally, Albihlal surveyed the genome-wide targets (ChIP-seq) of HSFA1b from Arabidopsis, and the significant motifs CArG, G-box, and FLY within HSFA1b peaks were discovered by using MEME-ChIP under both heat stress and non-stress conditions. The expression of AtHSFA6a is increased when exposed to salt, drought or abscisic acid (ABA) treatment, but is not significantly altered by heat and cold stresses. Further study showed that HSFA6b is involved in ABA-mediated regulons, and expression of HSFA6b improves tolerance to drought and salt mediated by ABA, and many stress-responsive genes are activated by AtHSFA6a (Hwang ; Huang ). Furthermore, HSFs are found to play a role in plant development. For example, an HSF from sunflower, HaHSFA9, is specifically expressed during embryogenesis and involved in embryogenesis developmental regulation (Almoguerac ; Díaz-Martín ).

Although the function of some HSFs in thermotolerance had been well characterized, their functions in other abiotic stresses remain largely unknown (Wu ). Many aspects need to be determined, including: Do HSFs only bind to the HSE to regulate the gene expression? What genes are regulated by HSFs in plants under salt stress conditions? Which physiological pathways are mediated by HSFs to improve the tolerance to abiotic stress? Answering these questions will clarify the function of HSFs in abiotic stress response.

Previous studies showed that the expression of AtHSFA7b is induced by heat stress, suggesting that it may play a role in heat tolerance (Yabuta, 2016). In addition, AtHSFA7b was found to be regulated by HsfA1d and HsfA1e, which are the key regulators in the HSF signaling network responding to environmental stress, suggesting that AtHSFA7b is also involved in environmental stress (Nishizawa-Yokoi ). In previous study, we found that AtHSFA7b was highly induced by salt treatment, indicating that it may be involved in salt stress tolerance. Therefore, we characterized the function of AtHSFA7b in salinity stress tolerance. In addition to binding to the HSE motif, we also identified a cis-acting element bound by AtHSFA7b using chromatin immunoprecipitation sequencing (ChIP-seq) and RNA sequencing (RNA-seq) combination assay. The physiological pathways regulated by AtHSFA7b were determined, and the genes involved in these physiological responses were further identified. Based on these results, we proposed a working model of AtHSFA7b in positively mediating salt stress tolerance.

Materials and methods

Plant materials and treatments

Seeds of Arabidopsis (Columbia ecotype) were seeded in a sand and peat mixtures (1:2 v/v) in a greenhouse with the conditions of a 10 h dark/10 h light photocycle, 70–75% relative humidity, and a stable temperature of 24 °C. The T-DNA insertion AtHSFA7b (At3g63350.1) mutant plants (SALK_152004) was obtained from the Arabidopsis Biological Resource Center (ABRC). Seedlings of Arabidopsis (4 weeks old) were watered with 150 mM NaCl solution for 0, 3, 6, 12, or 24 h, and were harvested for study.

Plasmid construction and genetic transformation

The intact coding sequence (CDS) of AtHSFA7b driven by the CaMV 35S promoter was cloned into pROK2 plant expression vector (Hilder ) generating the construct 35S:AtHSFA7b. Arabidopsis plants were genetically transformed with the construct 35S:AtHSFA7b with a flower-dipping method. The promoter of AtHSFA7b (1468 bp in length including 232 bp of 5′ untranslated region (UTR) sequence) was cloned into pCAMBIA1301 to replace the 35S promoter for driving a β-glucuronidase (GUS) gene (ProAtHSFA7b::GUS). The ProAtHSFA7b::AtHSFA7b construct was transformed into the AtHSFA7b mutant lines (SALK_152004) to generate the complementary transgenic lines (Chsf) to rescue the mutation of AtHSFA7b. The T3 homozygous lines of all the transgenic plants were used for analysis.

RNA-seq analysis

Four-week-old hsf(b) mutant and wild-type (WT) Arabidopsis plants (well watered) were treated with NaCl (watered with 150 mM NaCl solution) for 3 h and used for analysis. RNA-seq was performed to determine the differentially expressed genes (DEGs) between WT and hsf(b) Arabidopsis plants. Two independent biological replicates were carried out. The t-test P-values and the normalized values from these two samples were averaged. For selection of the DEGs, the criteria were set as normalized values <0.5 or >2 and t-test P-value<0.05.

MEME analysis

A Multiple EM for Motif Elicitation (MEME) study was carried out for determination of the sequences conserved among the promoters of the AtHSFA7b regulatory genes. The truncated promoters (2000 bp upstream from the translation initiation site) of 30 genes that were potentially regulated by AtHSFA7b according to RNA-Seq data were downloaded from the TAIR database (http://www.arabidopsis.org/) and subjected to a MEME assay (http://meme-suite.org/tools/meme).

Transactivation assay

The complete CDS and various truncated CDSs of AtHSFA7b were separately amplified using PCR (see Supplementary Table S1 at JXB online) and were cloned into pGBKT7 vector to fuse with GAL4 domain in frame (Clontech, CA, USA). These constructs were respectively transformed into yeast cells AH109, and cultured respectively on the media of SD/−Trp/X-α-Gal and SD/−Trp at 30 °C for 72–120 h. According to the prediction, the activation domain is located at the C terminus, whose main amino acids are F, W, L, and L. For determination of whether they are the main amino acids for the activation domain, the amino acids F, W, L, and L were all PCR mutated to H, and cloned into pGBKT7 for transactivation assay.

Subcellular localization study

The CDS (without the termination codon) of AtHSFA7b was fused with the CDS of green fluorescent protein (GFP) in-frame to form a HSF–GFP fusion protein, and this fusion protein was driven by the 35S promoter to generate vector 35S:AtHSFA7b-GFP. The 35S:GFP construct was used as a control. These two constructs were respectively transformed into epidermal cells of onion with biolistic delivery technology. The biolistic transformation was performed with a particle bombardment system (BiolistidB PDS-1000/He). In brief, 5 μl of DNA (1 μg μl−1), 50 μl of CaCl2 (2.5 mol l−1) and 20 μl of spermidine (0.1 mol l−1) were added one by one into 50 μl of gold particle solution (with a concentration of 60 mg ml−1). The particles were washed with absolute ethanol and re-suspended in 60 μl of absolute ethanol. Ten microliters of DNA/gold ethanol solution was loaded onto the macrocarrier center. Biolistic transformation was performed with the parameters as follows: the helium pressure of the rupture disk membranes was 1200 p.s.i., the distance between the rupture disk and macrocarrier was 3 cm, leaves were placed 9 cm away from the macrocarrier position, and the vacuum pressure was 26–28 in Hg. The transformed cells were examined by confocal microscopy at 20 h after transformation (LSM700, Zeiss, Jena, Germany). Syringe agroinfiltration was conducted on the tobacco cells to study the subcellular localization of AtHSFA7b. After transformation for 48 h, the transformed epidermal cells were examined by confocal microscopy.

ChIP-PCR and ChIP-seq analysis

Four-week-old Arabidopsis seedlings overexpressing GFP–AtHSFA7b were used for the ChIP analysis. The ChIP procedures were carried out following the description by Haring . In brief, 1% (v/v) formaldehyde was used to cross-link DNA and protein. The purified DNA–protein crosslinks (DPCs) were sonicated, and were immunoprecipitated with an anti-GFP antibody. The purified DPCs were also immunoprecipitated with anti-human influenza hemagglutinin (HA) antibody serving as the negative controls. The products of immunoprecipitation were precipitated using protein A agarose beads, and were incubated in 0.2 mol l−1 NaCl at 65 °C overnight for cross-link reversal. Immunoprecipitated DNA was isolated by chloroform extraction. ChIP-qPCR was performed to study the enrichment of the target DNA sequence. For the negative controls in ChIP, the promoters of Arabidopsis genes AtSOS2 and AtPOD4 were used. The CDS sequence of ACT7 and TUB2 were used as the internal references in ChIP-qPCR. All the primers used are shown as Supplementary Table S2. The fold enrichment of ChIP-qPCR was calculated according to the method described by Haring .

For ChIP-seq analysis, the reads from ChIP-sequencing were aligned to the genome of Arabidopsis from the TAIR database (http://www.arabidopsis.org/) by the Bowtie program (version 12.5) using a local alignment model with the default parameters. For discovery of the consensus DNA sequences from the AtHSFA7b binding peaks, these binding peaks were centered on their peak summits. MEME-ChIP, the online tool to discover de novo motifs, was used to identify the DNA motifs within peak sequences with the default parameters.

Yeast one-hybrid analysis

The CDS of AtHSFA7b was cloned into pGADT7-Rec2 (Clontech) as an effector. Three tandem copies of these studied motif sequence were cloned into pHIS2 vector as a reporter (see Supplemental Table S1). The binding of AtHSFA7b to the studied DNA sequences was determined with a yeast one-hybrid (Y1H) assay following the procedures of the Y1H kit (Clontech).

Electrophoretic mobility shift assay experiment

The intact coding sequence of AtHSFA7b was inserted into the vector pMAL (NEB) to fuse with the maltose-binding protein (MBP) CDS, generating vector MBP-AtHSFA7b, which was transformed into Escherichia coli ER2523. Recombinant proteins were affinity purified using Amylose Resin (New England Biolabs, cat. no. E8021S). The probes were synthesized and labeled with biotin. Different concentrations of non-labeled probe were added to the reactions for competition. An electrophoretic mobility shift assay (EMSA) was carried out with the Chemiluminescent EMSA kit (Beyotime, China). The sequences of probes are shown as Supplementary Table S3.

β-Glucuronidase and luciferase assays

For construction of reporter vectors, the different E-box sequences with three tandem copies were respectively fused with 46 bp minimal 35S promoter for driving a GUS reporter gene. The effector was the AtHSFA7b overexpression vector (35S:AtHSFA7b). The effector was transformed into tobacco together with each reporter. For normalizing the transformation efficiency, the 35S:Luc vector was also co-transformed. The activities of luciferase (Luc) and GUS were measured according to Lu and Qin . All the primers used are listed in Supplementary Table S1.

Transient transformation

The CDSs of these genes were separately cloned into pROK2 driven by the 35S promoter to construct the overexpression vector. To knockdown these genes, truncated cDNA sequences that were specific for each gene with inverted repeat were respectively cloned into the two sides of the CHSA intron in pFGC5941. All the primers used are shown as Supplementary Table S1.

Single colonies of Agrobacterium tumefaciens strain EHA105 harboring overexpression (OE), RNAi-silenced (knockdown, KD), or control (transformed with empty pROK2) vectors were cultured in LB medium with shaking to OD600 of 0.6, and A. tumefaciens cells were centrifuged at 2000 g for 5 min to harvest for transformation. Plant seedlings grown in a 1/2 MS solution (pH 5.8, 2.5% (w/v) sucrose) were used for transformation. The transformation was carried out following the description by Zang . Arabidopsis plants overexpressing the genes and with RNAi-silencing of the expression of the genes (knockdown) were generated, and empty pROK2 was transformed into Arabidopsis as a control. These plants were used for determination of the gene function in salt tolerance.

Salt stress tolerance assay

Seeds of Arabidopsis were sown on 1/2 MS medium containing 125 mM NaCl, and seeds sown on 1/2 MS medium were used as the control. After 7 d, the germination rates were measured. Arabidopsis seeds sown on 1/2 MS solid medium for 5 d were transferred to 1/2 MS solid medium supplying 150 mM NaCl for 10 d, and fresh weight and root length were measured. Seeds of Arabidopsis sown on 1/2 MS solid medium for 8 d were transferred to soil. After growth for 3 weeks, seedlings were watered with a solution of 200 mM NaCl for 10 d, and then imaged and analysed.

Physiological analysis

Measurement of chlorophyll contents was determined according to the description of Lightenthaler (1987). The superoxide dismutase (SOD) and peroxidase (POD) activities were determined according to Lagriffoul . The content of proline was determined following Bates . Malondialdehyde (MDA) was determined following Dhindsa . Stomatal apertures were determined according to Pei . Electrolyte leakage was determined following Campos . The contents of Na+ and K+ were measured according to Amini and Ehsanpour (2005). For reactive oxygen species (ROS) measurement, the plant material was ground into fine powders under liquid nitrogen, and ROS was measured using a ROS ELISA kit (Senbeijia Bio. Co., Nanjing, China) according to the manufacturer’s protocol. Three independent biological replications were performed. Water loss rate was measured following Mohammadkhani and Heidari (2008).

Quantitative RT-PCR analysis

Isolation of RNA from Arabidopsis plants was performed with Trizol reagent (Invitrogen). RNA was reverse transcribed into cDNA using oligo (dT) as template primers with the Primescript™ RT reagent kit (Takara) and diluted to 100 μl for RT-PCR amplification. The genes for actin 7 (At5G09810) and tubulin β-2 (At5G62690) served as the internal references. Real-time PCR was performed in a qTower 2.2 (Analytik Jena). The PCR reaction contained was as follows: SYBR Green Real-time PCR Master Mix (Toyobo, Japan) 10 μl, forward and reverse primers (0.5 μM each), cDNA template 2 μl and total volume 20 μl. Cycling parameters were as follows: denaturation at 94 °C 30 s, 45 cycles at 94 °C 12 s; 60 °C 30 s; 72 °C 40 s. Three independent biological repeats were carried out. The relative expression was calculated with the 2−ΔΔ method according to Livak & Schmittgen (2001). All relative expression data were log2 transformed. The primers used in PCR experiments are listed in Supplementary Table S4.

Statistical analysis

The statistical analysis was carried out using SPSS (version 18.0), and one-way analysis of variance (ANOVA) was used for analysis. Statistical significance was defined as P≤0.05.

Results

The expression profile of AtHSFA7b

To study the expression profile of AtHSFA7b, the GUS (β-glucuronidase) gene was used under the control of the promoter of AtHSFA7b to generate the plant expression vector ProHSFA7b:GUS. Arabidopsis plants transformed with ProHSFA7b:GUS were generated, and GUS staining was performed on the T3 homozygous plants (Fig. 1A). The expression of AtHSFA7b could be detected in the period from seed germination to flowering, and also could be detected in different tissues. In 5-week-old seedlings, the expression of AtHSFA7b was highest in roots compared with that in other tissues, including young leaves, pistils, sepals, and stamens, and was lowest in stems and siliques. Consistently, qRT-PCR assays also indicated that AtHSFA7b was most highly expressed in roots, moderately in leaves and flowers, but had very low expression in stems (Fig. 1B). These results indicated that AtHSFA7b has a tissue-specific expression pattern.

Fig. 1.

The expression pattern of AtHSFA7b. (A) The promoter of AtHSFA7b was fused with a GUS gene (ProHSFA7b:GUS) and transformed into Arabidopsis; the expression of AtHSFA7b was determined using histochemical GUS staining and activity. (1) Seeds; (2) 3-day-old seedlings; (3) 5-day-old seedlings; (4) 7-day-old seedlings; (5) 20-day-old seedlings; (6) rosette leaves; (7) flowers; (8) sepals; (9) siliques; (10) 30-day-old seedlings. (B) The transcripts of AtHSFA7b in different tissues of Arabidopsis plant. Four-week-old Arabidopsis plants were determined using qRT-PCR. (C, D) The expression of AtHSFA7b in leaves (C) or in roots (D) in response to salt stress. Well-watered Arabidopsis plants were harvested at each stress time point and used as controls. At each treatment time point, the relative expression level of AtHSFA7b was calculated as its expression level under stress condition divided by its expression under normal condition at this time point. All expression values were log2 transformed. (E) GUS staining and activity analysis of ProHSFA7b:GUS transformed plants under salt stress conditions. The same weight of fresh leaves or roots under normal or NaCl stress condition were used for GUS activity analysis. (This figure is available in color at JXB online.)

To characterize the expression of AtHSFA7b under salt stress conditions, qRT-PCR was conducted. In roots and leaves, the transcripts of AtHSFA7b were significantly increased after salt stress for 3 h, reached the highest level at 6 h of stress, and then decreased after 6 h (Fig. 1C, D). Consistently, GUS staining and activity analysis of the Arabidopsis plants expressing ProHSFA7b:GUS also showed that AtHSFA7b was induced by salt treatment in both roots and leaves (Fig. 1E), suggesting that AtHSFA7b is involved in the salinity stress response.

A phylogenetic tree was constructed using AtHSFA7b and 22 other HSF proteins from Arabidopsis, and the results showed that AtHSFA7b protein has a close genetic distance to AtHSFA7a, AtHSFA3, and AtHSFA6b (see Supplementary Fig. S1).

AtHSFA7b is a transcriptional activator

For characterization of the transcriptional activation activity of AtHSFA7b, the CDS of AtHSFA7b was serially deleted, respectively cloned into a pGBKT7 vector to fuse with a GAL4 domain in-frame, and transformed into yeast two-hybrid (Y2H) yeast cells. The yeast cells transformed with the intact CDS of AtHSFA7b can grow well on medium of SD/−Trp, and the colonies appeared blue on the SD/−Trp medium supplied with X-α-Gal, indicating that AtHSFA7b is a transcriptional activator. Additionally, the minimal truncated CDS of AtHSFA7b with transcriptional activation activity is located at amino acid residues from 213 to 282 in AtHSFA7b, suggesting that this region contains the transcriptional activation domain (Fig. 2A). In addition, the predicted activation domain is located at the C-terminus with the amino acids F, W, L, and L (Fig. 2A), and when they were mutated to H, the transcriptional activation activity of AtHSFA7b was lost, indicating that the main amino acids in the transactivation motif should be F, W, L, and L.

Fig. 2.

Analysis of transactivation and subcellular localization of AtHSFA7b. (A) Transactivation analysis of AtHSFA7b. Different truncated coding sequences (CDS) of AtHSFA7b, together with the truncated CDS whose amino acids F, W, L, and L were mutated to H, were respectively fused with GAL4 domain in-frame and cloned into pGBKT7. These constructs were transformed into Y2H yeast strain for transactivation assay. The yeast transformants were grown on the medium SD/−Trp (transformation control) and the medium SD/−Trp/X-α-Gal, respectively. (B) AtHSFA7b was fused with GFP gene driven by 35S promoter (35S:AtHSFA7b-GFP) and was transformed separately into epidermal cells of tobacco plants (1–6) and onion (7–14). The 35S:GFP transformed onion epidermal cells served as controls. In the onion epidermal cells, the nuclei were stained with 4′,6-diamidino-2-phenylindole (DAPI). (3, 6, 7, 11) GFP fluorescence; (2, 5, 8, 12) bright field; (9, 13) DAPI fluorescence; (1, 4, 10, 14) merged bright field and fluorescence images. (This figure is available in color at JXB online.)

For subcellular location analysis, 35S:AtHSFA7b-GFP transformed tobacco epidermal cells and onion epidermal cells both showed GFP fluorescence in the nuclei, while GFP fluorescence in the 35S:GFP transformed cells was distributed uniformly (Fig. 2B). Therefore, AtHSFA7b is located in nuclei. Taken together, AtHSFA7b is a nuclear protein with transcriptional activation activity, demonstrating that it has the characteristics of a transcription factor.

Transgenic plant overexpression of AtHSFA7b and complementation of AtHSFA7b mutant Arabidopsis plants

Transgenic plants for overexpression of AtHSFA7b were generated; at the same time, the construct harboring AtHSFA7b gene controlled by its own promoter was generated and transformed into the mutant line (SALK_152004) (hsf) to generate the complementary lines (Chsf). The expression of AtHSFA7b in OE lines, mutant plants (hsf), WT and complemented mutant lines (Chsf) were determined using qRT-PCR. The transcript levels of AtHSFA7b were highly elevated in all the transgenic OE lines, but were significantly reduced in the hsf mutant line, and were expressed similarly in the WT plants and all the Chsf complementation lines (Fig. 3A), indicating that the OE, mutant hsf, and complementary mutant Chsf plants were all successfully generated.

Fig. 3.

AtHSFA7b confers salt tolerance to Arabidopsis plants. (A) The expression of AtHSFA7b in different types of plant lines. WT, wild-type Arabidopsis plants; OE-1–5: the independent transgenic lines overexpressing AtHSFA7b; hsf(a)–(e), the AtHSFA7b mutation plants from the same Arabidopsis mutation line (SALK_152004); Chsf-1–5, the independent complementary mutants of AtHSFA7b, which were generated by transformation of hsf plants with AtHSFA7b under the control of its own promoter. All expression values were log2 transformed. (B) Seed germination rate analysis. Seeds were plated on 1/2 MS solid medium (normal) or on 1/2 MS containing 150 mM NaCl (NaCl). After growth for 1 week, the germination rates were compared. (C) The growth phenotype of different types of plant lines in culture medium under salinity or normal conditions. Seeds were plated on 1/2 MS medium (normal) for 1 week, and then plated on 1/2 MS or 1/2 MS medium containing 150 mM NaCl (NaCl) for 1 week to compare the growth phenotype. Three independent biological experiments were performed. (D) Four-week-old seedlings in soil were watered on their roots with fresh water or 150 mM NaCl solution for 5 d. Comparison was made of plant growth in soils under salinity or normal conditions. (E) Four-week-old seedlings were watered with fresh water or 150 mM NaCl solution on their roots for 5 d. Comparison of contents of chlorophyll among WT, OE, hsf and Chsf under salt stress or normal conditions. (This figure is available in color at JXB online.)

Expression of AtHSFA7b improves tolerance to salt and heat stresses

The expression of AtHSFA7b is induced by salt stress, demonstrating that it may play a role in salinity stress. Therefore, we investigated its function in salt stress, and four different plant lines were studied, namely WT, three independent OE lines (OE-1, OE-2, and OE-3), two mutant plants (hsf(b) and hsf(c) that are from the same line SALK_152004), and three independent complementary mutant lines (Chsf-1, Chsf-2, and Chsf-3). There was no difference among all the studied plants in plant growth, root length, fresh weight, and seed germination rates under normal growth conditions (Fig. 3B, C). However, when exposed to salt stress, all OE lines displayed increased fresh weight, root length, and seed germination rate compared with WT and Chsf complementary lines, which were similar for these parameters; the mutant plants showed the lowest fresh weight, root length, and seed germination rate (Fig. 3B, C). To further characterize the salt tolerance of AtHSFA7b, the plants grown in soil were studied. Consistently, all the studied plants were similar in growth phenotype without stress. Therefore, under normal conditions, overexpression or mutation of AtHSFA7b does not affect plant growth. However, when exposed to salt stress, the Chsf complementary lines displayed similar growth phenotypes to the WT Arabidopsis. Three OE lines showed substantially improved salt tolerance, while the hsf plants showed highly decreased salt tolerance when compared with the Chsf and WT plants (Fig. 3D, E). All the studied plants had similar chlorophyll contents under normal growth condition (Fig. 3D, E). The contents of chlorophyll were decreased in all the studied plants under salt stress conditions. However, OE lines still had a higher chlorophyll content than Chsf complementary lines and WT plants, and hsf plants had the lowest chlorophyll contents (Fig. 3D, E). Taken together, AtHSFA7b positively regulates salt tolerance, making it deserving of further study.

The heat stress tolerance of AtHSFA7b was also determined. The 5-week-old Arabidopsis (including OE, WT, mutant plants hsf(b), and complementary lines Chsf) were treated with thermal stress at 44 °C for 120 min, followed by recovery for 7 d. Their growth phenotype, MDA, and chlorophyll contents were assessed. In the absence of heat treatment, the studied plants all had similar growth phenotypes and contents of MDA and chlorophyll (see Supplementary Fig. S2A–D). However, after heat stress, the OE lines showed relatively slight injury, with the highest chlorophyll content and lowest MDA levels, while there was moderate injury in Chsf complementary lines and WT plants. The hsf mutant plants suffered serious injury, having the lowest chlorophyll content and the highest MDA level (Supplementary Fig. S2A–D). Therefore, AtHSFA7b could also confer thermal tolerance to plants.

RNA-seq studies was performed to determine the DEGs between WT and hsf plants under salt stress conditions. In total, 1644 DEGs were identified; among them, 1093 and 551 genes were found to be induced and inhibited by AtHSFA7b, respectively (see Supplementary Table S5). Ten DEGs were selected randomly to validate the reliability of RNA-seq data using qRT-PCR. The results of qRT-PCR and RNA-seq are generally consistent in the expression patterns of these 10 DEGs (Supplementary Fig. S3A), indicating that RNA-seq data were reliable. Gene ontology (GO) classification was performed for the DEGs. In the biological process category, the DEGs were highly expressed in the categories of immune system process, rhythmic process, and cell killing (Supplementary Fig. S3B). In the molecular function category, DEGs were highly accumulated in channel regulator activity, nutrition reservoir activity, antioxidant activity, and electron carrier activity (Supplementary Fig. S3B).

Physiological analysis of salt tolerance mediated by AtHSFA7b

To determine the salt tolerance mechanism mediated by AtHSFA7b at the physiological level, physiological studies were performed. Under normal conditions, all the plants had similar electrolyte leakage rates (Fig. 4A). When exposed to salt stress, the hsf mutant plants had higher electrolyte leakage than Chsf and WT plants, which shared similar electrolyte leakage; however, all OE plants displayed the lowest electrolyte leakage among the studied plants (Fig. 4A).

Fig. 4.

The stress-related physiological changes mediated by AtHSFA7b. (A) Comparison of electrolyte leakage. (B) Analysis of soluble sugar contents. (C) Proline content comparison. (D) The expression of the P5CS genes in WT, OE, hsf, and Chsf lines. (E) Water loss rate assay. 0–7 h: The detached leaves were exposed to air in a clean bench. After exposure to air from 0.5 to 7 h, the leaves were weighed at different time points. (F) Determination of stomatal aperture. (G) The ratios of width/length of stomatal aperture. (H) The expression of stomatal aperture control-related genes in WT, OE, hsf, and Chsf lines. All expression values were log2 transformed. *Significant difference compared with WT (P<0.05).

Both proline and soluble sugars are osmolytes in plants for resisting adverse environments. To study whether AtHSFA7b regulates the biosynthesis of osmolytes to improve salt tolerance, the contents of proline and soluble sugars were determined. Under normal conditions, the contents of proline and soluble sugars were similar in all the studied plant lines, but were increased in all the studied lines when exposed to NaCl treatment. In addition, the contents of proline and soluble sugars were significantly induced in the OE lines, while they were significantly reduced in hsf plants as compared with the Chsf and WT plants under salt stress conditions (Fig. 4B, C). Two proline biosynthesis genes, AtP5CS1 and AtP5CS2 (encoding Δ1-pyrroline-5-carboxylate synthases), were studied. All the studied plants had similar transcript levels of AtP5CS1 and AtP5CS2 under normal conditions. When suffering from salt stress, all OE lines had the highest transcript levels of AtP5CS1 and AtP5CS2, and the hsf plants had the lowest expression levels of AtP5CS1 and AtP5CS2 in comparison with Chsf and WT plants; WT and Chsf plants shared similar expression levels of AtP5CS1 and AtP5CS2 under conditions of salt stress or normal conditions (Fig. 4D).

As AtHSFA7b can improve salt tolerance, we further studied whether the ability to conserve water in the transgenic plants is altered by AtHSFA7b. Water loss rates were first studied. The results showed that Chsf lines and WT plants shared similar water loss rates (Fig. 4E). With increase in time, water loss rates in the OE lines were highly decreased, while they were highly elevated in hsf mutant plants in comparison with the Chsf and WT plants (Fig. 4E). To determine whether the changed water loss rate was due to altered stomatal opening, the stomatal apertures were studied. Under normal conditions, all the studied plants showed similar ratios of stomatal width/length. However, under salt stress conditions, WT and Chsf lines share similar ratios of width/length; however, hsf mutant plants displayed the highest width/length ratios and opening of stomata among the studied plants, and OE lines displayed reduced opening of stomata when compared with that of the WT and Chsf lines (Fig. 4F, G). The expression of genes controlling changes in stomatal aperture was studied, including AtMYB61 (myb domain protein 61; At1G09540), AtHPP2C5 (phosphatase 2C5; At2G40180), and RHC1 (RESISTANT TO HIGH CO2; At4G22790). Under salt stress conditions, the expression of AtHPP2C5 and RHC1 was induced significantly in the OE lines, and was decreased significantly in hsf plants as compared with Chsf and WT plants. However, AtMYB61 was not significantly regulated by AtHSFA7b (Fig. 4H). In the WT and Chsf plants, the expression of these genes was similar (Fig. 4H). Therefore, AtHSFA7b could induce ATHPP2C5 and RHC1 expression to control the change of stomatal aperture, but does not regulate the expression of AtMYB61.

The content of K+ or Na+ was measured in roots and leaves of plants. The contents of Na+ or K+ and K+/Na+ ratios were similar in both roots and leaves among all the studied lines under normal conditions (Fig. 5A–D). When exposed to salt stress, all the plants displayed decreased K+ level in both roots and leaves; however, OE lines still retained the highest K+ level, the hsf mutant showed the lowest K+ level, and Chsf and WT plants had similar K+ content (Fig. 5A, B). In addition, all the plants had increased Na+ level when suffering from salt stress; however, hsf mutant plants had the highest Na+ contents, Chsf and WT had similar Na+ contents, and OE plants had a lower Na+ content than both Chsf and WT (Fig. 5C, D). In both leaves and roots, OE lines retained the highest ratio of K+/Na+ among all studied lines when exposed to salt stress (Fig. 5E, F).

Fig. 5.

Cellular homeostasis of Na+ and K+ regulated by AtHSFA7b. Arabidopsis seedlings in soil were watered with NaCl solution (150 mM) for 7 d, and used for study. The well-watered plants served as the controls. (A) K+ contents in leaf tissue; (B) K+ contents in root tissue; (C) Na+ contents in leaf tissue; (D) Na+ contents in root tissue; (E) ratios of K+/Na+ in leaf tissue; (F) K+/Na+ ratios in root tissue. (G) The expression of K+ or Na+ homeostasis-related genes in WT, OE, hsf, and Chsf plants. All expression values were log2 transformed. *Significant difference compared with WT (P<0.05).

The genes involved in transport of K+ or Na+, including salt overly sensitive (SOS) (SOS1 to SOS3) and Na+ (K+)/H+ transporter (NHX1 to NHX6), were studied (Fig. 5G). The transcripts of NHX4 and NHX5 were not changed in all the studied plants under salt stress conditions. However, the other seven genes, including NHX1, NHX2, NHX3, NHX6, SOS1, SOS2, and SOS3, were all induced significantly in OE plants, and were highly decreased in hsf mutant plants as compared with WT and Chsf (Fig. 5G). These studies suggested AtHSFA7b induces the expression of NHX1, NHX2, NHX3, NHX6, SOS1, SOS2, and SOS3 to reduce ion toxicity caused by salt stress.

Determination of ROS content showed that all the studied plants shared similar ROS and MDA levels in the absence of salt stress (see Supplementary Fig. S4D, E). However, all the studied plants showed increased ROS and MDA contents under salt stress conditions. In addition, hsf mutant plants had the highest contents of ROS and MDA, followed by the WT plants, which showed similar ROS levels to the Chsf, while all the OE plants had the lowest contents of MDA and ROS (Supplementary Fig. S4D, E). To study whether the increased lipid peroxidation and ROS were caused by altered antioxidant activities, the activities of glutathione-S-transferase (GST), peroxidase (POD) and superoxide dismutase (SOD) were studied. Under normal conditions, all the studied plants had similar SOD, POD, and GST activities (Supplementary Fig. S4A–C). Under salt stress conditions, all OE plants had the highest SOD, POD, and GST activities, and Chsf lines and WT also had significantly higher activities of SOD, POD, and GST than those of hsf mutant plants (Supplementary Fig. S4A–C). RNA-seq data showed that seven GST genes were induced by AtHSFA7b (Supplementary Table S5), indicating that AtHSFA7b could induce GST gene expression to increase GST activity. The expression of the POD and SOD genes was studied using qRT-PCR. Consistent with the activities of POD and SOD, all the studied plants had similar expression levels of POD and SOD genes under normal conditions (Supplementary Fig. S4F, G). However, when exposed to salt, the expression of POD and SOD genes were all lowest in the hsf plants, moderate in Chsf lines and WT plants, and highest in the OE lines (Supplementary Fig. S4F, G). All the above results together indicated that AtHSFA7b could activate the expression of POD, SOD, and GST genes to improve ROS scavenging activity.

AtHSFA7b preferentially binds to an E-box-like elements

ChIP-seq was performed to identify the genome-wide targets of AtHSFA7b. The input DNA (sonicated DNA before immunoprecipitation) was used as the negative control to eliminate non-specific peaks, and 4748 specific peaks potentially bound by AtHSFA7b were identified. The distribution of peak sequences in the genome of Arabidopsis was studied. In total, 36.12% of the peaks were distributed in the promoter regions, 60% of the peaks were distributed in distal intergenic regions, and the other peaks were distributed in the regions of UTR, exon or intron (Fig. 6A). The genes associated with these peaks were identified, and 2101 genes were found to be closely associated with the ChIP-seq peaks (see Supplementary Table S6). Among the 2101 genes, there are 1955 genes annotated by the GO annotation. GO classification of these genes showed that the terms protein binding, transcription factor, receptor, nutrient reservoir, and guanyl-nucleotide exchange factor were highly enriched (Fig. 6B). Motif discovery analysis was performed using MEME-ChIP to determine the conserved DNA sequences. There were 18 conserved motifs identified (Fig. 6C), indicating that AtHSFA7b may bind to these.

Fig. 6.

ChIP-seq analysis of the binding sites of AtHSFA7b. (A) Analysis of AtHSFA7b-bound genomic regions. (B) GO analysis of the target genes associated with the peaks from ChIP-seq. (C) The conserved sequences identified using MEME-ChIP. (D) MEME analysis of the conserved sequences present in the promoters of target genes of AtHSFA7b identified by RNA-seq. (E) Identification of the bindings of AtHSFA7b to the peak sequences from ChIP-seq that contain some E-box sequences using ChIP-qPCR. Fold enrichment was calculated as the abundance of aim DNA relative to internal reference in ChIP+ divided by the abundance of aim DNA relative to internal reference in input. The fold enrichment value was log2 transformed. (This figure is available in color at JXB online.)

To determine the motifs that were bound by AtHSFA7b to regulate gene expression, 30 genes that were induced highly by AtHSFA7b were selected for study, and the promoters of these genes (−2000 to 0 bp) were retrieved and subjected to MEME analysis. We found that a conserved motif identified in MEME-ChIP was also identified by MEME analysis of these promoter regions, which contains some types of E-box sequences (Fig. 6D), and is similar to the conserved sequences identified by MEME-ChIP (Fig. 6C). ChIP-qPCR was performed to confirm whether some of the peak sequences from ChIP-seq that contain E-box sequences were really bound by AtHSFA7b, and four peak sequences with E-box sequences were randomly selected for study. The primers used for ChIP-qPCR are shown as Supplementary Table S2. ChIP-qPCR showed that these peak sequences were all bound by AtHSFA7b (Fig. 6E), indicating the reliability of the peak prediction result.

To determine the exact E-box sequences bound by AtHSFA7b, a yeast Y1H assay was performed to study the binding of all E-box sequences to AtHSFA7b. AtHSFA7b can only bind to five types of E-box motif, including 5′-CACGTG-3′, 5′-CATTTG-3′, 5′-CAATTG-3′, 5′-CAGCTG-3′, and 5′-CATATG-3′; the other E-box motifs are not bound by AtHSFA7b (Fig. 7A). Therefore, these five types of E-box sequences, namely 5′-CACGTG-3′, 5′-CATTTG-3′, 5′-CAATTG-3′, 5′-CAGCTG-3′, and 5′-CATATG-3′, were termed E-box-like motifs. To further validate the above results, reporter constructs were built by using different E-box motifs fused with a 35S minimal promoter (−46 bp) for driving the GUS report gene. Each reporter and the effector (35S:AtHSFA7b) were transiently co-transformed into tobacco. Consistently, the ratios of GUS/Luc showed that AtHSFA7b binds to all E-box-like motifs, but failed to bind to other E-box sequences (Fig. 7B). EMSA was further conducted with all the E-box-like motif as the probes, including 5′-CACGTG-3′, 5′-CATTTG-3′, 5′-CAATTG-3′, 5′-CAGCTG-3′, and 5′-CATATG-3′. The EMSA results indicated that protein–DNA complex bands were observed only when the probes of E-box-like sequences interacted with AtHSFA7b protein. In addition, with unlabeled competitor probes increasing, the bound complex signal intensities were decreased gradually (Fig. 7C), confirming the results that AtHSFA7b binds to these sequences. Therefore, AtHSFA7b only binds to some E-box sequences, i.e. E-box-like motifs, including 5′-CACGTG-3′, 5′-CATTTG-3′, 5′-CAATTG-3′, 5′-CAGCTG-3′, and 5′-CATATG-3′.

Fig. 7.

The AtHSFA7b protein binds to an E-box-like motifs. (A) Y1H study of the binding of AtHSFA7b to E-box-like motifs. (B) The interaction between AtHSFA7b and E-box-like motifs in tobacco leaves. The effector and reporter constructs were shown at the top of the figure. (C) EMSA study on the binding of AtHSFA7b to the E-box-like motifs; 10, 50, and 100-fold excess of unlabeled probes were respectively added as the competitors. (D) Analysis of enrichment of truncated promoters using ChIP-qPCR. The truncated promoters containing the E-box-like motifs of the genes (including At5G32825, At4G05632, At2G12210, At3G30630, At2G40180, AtSOD3, AtSOD4, AtSOD5, AtPOD2, AtPOD10, AtSOS1, AtNHX3, and AtNHX6) were analysed. Fold enrichment was calculated as the abundance of aim DNA relative to internal reference in ChIP+ divided by the abundance of aim DNA relative to internal reference in input. The fold enrichment values were log2 transformed. (E) Agarose gel electrophoresis analysis of ChIP-PCR products. ACT7, TUB2: the internal references used in ChIP-qPCR. AtSOS2, AtPOD4: the negative controls that do not contain any E-box-like or HSE motifs in their promoter regions; a, input, the chromatins without immunoprecipitation; b, ChIP+, sonicated chromatins immunoprecipitated using an anti-GFP antibody; c, ChIP−, sonicated chromatins immunoprecipitated using anti-HA antibody; M, DNA marker (DL-2000).

To further determine whether AtHSFA7b actually binds to these E-box-like motifs in Arabidopsis, ChIP-PCR was carried out. Chromatin immunoprecipitated with anti-HA antibody served as the negative control. The genes regulated by AtHSFA7b (from the results of RNA-seq and qRT-PCR), and whose promoter regions contain E-box motifs, were randomly selected for ChIP-PCR analysis. ChIP-qPCR indicated that the truncated promoter sequences containing the E-box-like motifs were all significantly enriched (including At5G32825, At4G05632, At2G12210, At3G30630, AtSOD3, AtSOD4, AtSOD5, AtPOD2, AtPOD10, AtSOS1, At2G40180, AtNHX3, and AtNHX6), and two negative controls (AtSOS2 and AtPOD4) that do not contain any E-box-like or HSE motifs were also not enriched (Fig. 7D). The PCR products of ChIP was also analysed on agarose gel electrophoresis, and the results are consistent with ChIP-qPCR (Fig. 7E). The above studies together suggested HSFA7b can regulate the expression of genes by binding to E-box-like motifs in Arabidopsis under natural conditions.

Some target genes of AtHSFA7b are involved in tolerance to salinity stress

RNA-seq results showed that some gene families were highly regulated by AtHSFA7b, implying that these gene families might play roles in salinity tolerance. For instance, among these DEGs, 31 genes involved in cellulose synthesis were significantly down-regulated by AtHSFA7b, including cellulose synthase (CESA), cellulose synthase-like gene (CSLG), and cellulose synthase-like A (CSLA) family genes; certain transcription factors, such as basic helix–loop–helix (bHLH), NAC (NAM, ATAF, and CUC), WRKY, MYB and ZFP (zinc finger protein) families, were significantly induced by AtHSFA7b (see Supplementary Table S5). To determine whether these DEGs were involved in salt tolerance, we used a transient transformation system.

Vectors for overexpression or RNA interference (RNAi) silencing of these genes were constructed, and were separately transiently transformed into Arabidopsis plants for a gain- and loss-of-function analysis; at the same time, Arabidopsis transformed with pROK2 empty vector served as the control (Con). The plants for transient overexpression (transformed with 35S gene) or transient RNAi silencing (KD, transformed with pFGC5941) of the target genes, together with Con samples, were treated with salinity or under normal conditions. To determine the transcripts of transgenes in these transiently genetic transformed plants, qRT-PCR was performed. All the transformed plants showed increased expression of transgene in OE plants, but displayed a highly reduced expression level in KD plants in comparison with the Con samples (Fig. 8A, B), suggesting that these transgenes had been successfully overexpressed or knocked down in the transiently transformed plants, and therefore can be used for further study.

Fig. 8.

Salt stress tolerance assay of the target genes of AtHSFA7b. (A, B) Detection of the transcripts of AtHSFA7b in OE, Con and KD plants using qRT-PCR under normal growth condition (A) and salinity conditions (B). All expression values were log2 transformed. Arabidopsis plants transiently overexpressing the genes (OE), or with transient RNAi silencing the genes (knockdown, KD) were generated, and empty pROK2 was transiently transformed into Arabidopsis as control. (C) MDA content assay. AtCESA4: At5G44030; AtCESA8: At4G18780; AtCSLG1: At4G24000; AtCSLG2: At4G24010; AtCSLG3: AT4G23990; AtCSLG9: At5G03760; bHLH: At5G51780; NAC061: At3G44350; NAC036: At2G17040; NAC090: At5G22380; WRKY38: At5G22570; ZFP2: At5G57520; MYB63: AT1G79180. (D) Electrolyte leakage assay. (E) Analysis of ROS levels. *Significant difference compared with WT (P<0.05).

Of the studied cellulose synthase-related genes, except for CEA4 and CSLG3 that were not involved in salt tolerance, all were sensitive to salinity stress. For instance, compared with the Con plants, knockdown of CESA8, CSLG1, CSLG2, and CSLA9 significantly decreased the MDA levels, ROS levels, and electrolyte leakage rates, while overexpression of these genes could significantly increase the MDA levels, ROS levels, and electrolyte leakage rates (Fig. 8C, D, E), indicating that these genes negatively regulate salt tolerance.

Some transcription factors regulated by AtHSFA7b are found to play positive roles in salt tolerance. Except for MYB63, overexpression of the transcription factors, including bHLH, NAC061, NAC036, NAC090, WRKY38, and ZFP2, reduced the MDA levels, ROS levels, and electrolyte leakage rates compared with those in the Con plants; by contrast, the plants with RNAi-silencing of these TFs showed increased MDA levels, ROS levels, and electrolyte leakage rates (Fig. 8C, D, E), indicating that these TFs positively regulate salt tolerance.

Discussion

AtHSFA7b is involved in tolerance to salinity and heat

Like other HSFs, AtHSFa7b was found to confer significant thermal tolerance to plants (see Supplementary Fig. S2). However, the expression of AtHSFA7b is significantly increased under salt stress conditions (Fig. 1C–E), indicating a role in salinity tolerance. Therefore, we further determined whether it is involved in salt tolerance. The results indicated that expression of AtHSFA7b can significantly improve salinity tolerance, while knockout of AtHSFA7b in plants caused sensitivity to salinity stress (Fig. 3A–E), suggesting that AtHSFA7b positively regulates both heat and salinity tolerance.

AtHSFA7b has transactivation activity and preferentially binds to E-box-like elements

In this study, we found that AtHSFA7b is a nuclear protein, and has transcription activation activity, with the transactivation domain at its C-terminus (Fig. 2A), indicating that AtHSFA7b serves as a transcription factor to activate the expression of genes. Previous studies showed that HSFs are able to bind to HSEs (5′-AGAAnnTTCT-3′) (Akerfelt ). In the present study, MEME-ChIP (Fig. 6C) combined with MEME analysis (Fig. 6D) showed that AtHSFA7b preferentially binds to an E-box-like motif, although it also binds to HSE motifs (see Supplementary Fig. S5). We used a Y1H assay (Fig. 7A) and EMSA study (Fig. 7C) to further determine the accuracy of the sequences of E-box-like motifs. In addition, co-expression of 35S:AtHSFA7b and the E-box-like motifs driving a GUS gene showed that AtHSFA7b could bind to E-box-like motifs to activate GUS expression in plants (Fig. 7B), further suggesting that AtHSFA7b binds to E-box-like motifs to active the expression of genes. Furthermore, the ChIP results also confirmed that AtHSFA7b binds to E-box-like motifs to regulate gene expression (Fig. 7D, E). Therefore, the E-box-like motifs are quite important for gene expression regulated by AtHSFA7b under salinity stress conditions.

The physiological changes mediated by AtHSFA7b in response to salt stress

Soluble sugars are key osmolytes for osmotic adjustment in plants and a major indicator for plant tolerance to stress (Xiong ). Our study showed that the soluble sugar content was positively regulated by AtHSFA7b (Fig. 4B). Proline is one of the key osmolytes that stabilizes membranes and macromolecules in plant cells under osmotic conditions (Mahajan and Tuteja, 2005; Ijaz ). In Arabidopsis, there are two proline biosynthesis genes, AtP5CS1 and AtP5CS2 (Székely ). AtHSFA7b induces the expression of both AtP5CS1 and AtP5CS2 (Fig. 4D). The proline content also increases in response to the expression of AtHSFA7b in plants (Fig. 4C). Therefore, AtHSFA7b induces the expression of AtP5CS1 and AtP5CS2, which leads to increased proline levels to improve salt stress tolerance. In both halophytes and glycophytes, organic osmolytes, such as proline and soluble sugars, play important roles in improving osmotic tolerance (Wang ); therefore, the increase in both proline and soluble sugar levels mediated by AtHSFA7b suggested that AtHSFA7b could adjust the osmotic potential to improve salt tolerance.

Stomata mediate transpiration and CO2 influx from the atmosphere (Kim ). In plants, most of the water loss occurs through transpiration mediated by the stomata, and therefore control of the stomatal aperture is important for plant stress tolerance. As the water loss rate was correlated negatively with the expression level of AtHSFA7b (Fig. 4E), we further determined whether AtHSFA7b could control changes of the stomatal aperture. The results showed that the expression of AtHSFA7b was negatively associated with increased stomatal aperture (Fig. 4F, G). At the same time, the genes that control the change of stomatal aperture, namely AtHPP2C5 (At2G40180) and RHC1 (At4G22790), were induced by AtHSFA7b (Fig. 4H). Both AtHPP2C5 and RHC1 are involved in stomatal aperture control (Brock ; Tian ), and their expression patterns were consistent with the changes of stomatal apertures in OE, WT, Chsf, and hsf lines (Fig. 4G–H). These results suggested that AtHSFA7b controls changes of stomatal aperture through regulating AtHPP2C5 and RHC1 expression. Therefore, when exposed to stress conditions, AtHSFA7b could reduce the stomatal apertures by inducing the expression of AtHPP2C5 and RHC1, ultimately leading to a decreased rate of water loss to improve salt tolerance.

Plants generate high levels of ROS when they encounter adverse environments. ROS play a dual role, acting as signaling molecules at a low level, but at excess levels, they cause oxidative stress and damage macromolecules such as DNA, RNA, proteins, and lipids (Mittler, 2017). Therefore, regulation of ROS at suitable levels is quite important for plants under abiotic stress conditions (Zhang ). Our results showed that AtHSFA7b can induce the expression of genes including SOD, POD, and GST (see Supplementary Fig. S4F, G; Supplementary Table S5); at the same time, their activities were also correspondingly increased (Supplementary Fig. S4F, G), accompanied by decreased ROS levels (Supplementary Fig. S4D). In addition, GO classification analysis showed that genes involved in antioxidant activity were highly expressed (Supplementary Fig. S3B). These results together suggested that AtHSFA7b could induce ROS scavenging-related genes, including GSTs, SODs, and PODs, to increase the activities of GST, POD and SOD, resulting in decreased accumulation of ROS and finally leading to improved salt tolerance.

Plant cells alleviate excess cytosolic Na+ by compartmentalizing Na+ into vacuoles, and NHX proteins play important roles in this process. NHX proteins catalyse the electroneutral exchange of Na+ and/or K+ for H+ to direct the movement of Na+ or K+ in exchange for H+ (Yamaguchi ; Bassil ). In addition, overexpression of NHX can improve tolerance to salt stress (Mishra ). The SOS pathway, including SOS1, SOS2 and SOS3, mediates cellular signaling to maintain ion homeostasis (Ji ). Our results showed that the transcripts of AtNHX1, AtNHX2, AtNHX3, AtNHX6, AtSOS1, AtSOS2, and AtSOS3 were all positively regulated by AtHSFA7b (Fig. 5G). In addition, when exposed to salt stress, K+ loss was decreased and the accumulation of Na+ was also reduced with the increased expression of AtHSFA7b, leading to the maintenance of higher K+/Na+ ratios (Fig 5A–F). Therefore, to maintain cellular homeostasis, AtHSFA7b can compartmentalize excess cytosolic Na+ and reduce K+ loss through inducing the expression of the genes of the SOS signaling pathway and NHX family, which contribute to the improved tolerance to salt mediated by AtHSFA7b.

The cellulose biosynthetic process plays a role in salinity tolerance

The plant cell wall is the front line to protect the cell against adverse environments, and is composed of polysaccharides. These can be classified into pectins, hemicelluloses, and cellulose. Cellulose is the main cell wall component, and is synthesized by cellulose synthase (CesA) enzymes. Cellulose synthase (CESA) genes play roles in the synthesis of primary and secondary cell wall (Zhang ). Secondary walls have a large impact on plant adaptation to the environment, and some CESA genes and cellulose synthase-like genes (CSL) are involved in salt tolerance (Zhu ; Zhu ; Wang ; Zhang ; Li ; Zhang ). Our RNA-Seq data indicated that there were 31 cellulose synthase family genes down-regulated by AtHSFA7b (see Supplementary Table S5), indicating that they might be involved in AtHSFA7b-mediated salt tolerance. To determine whether AtHSFA7b mediates salt tolerance by down-regulation of these genes, CESA4, CESA8, CSLG1, CSLG2, CSLG3, and CSLA9 were randomly selected for salt tolerance study. A loss- and gain-of-function study using transient transformation indicated that except for CESA4 and CSLG3, the other four genes all negatively regulate salt stress tolerance (Fig. 8C–D). The results together indicated that cellulose synthesis-related genes were involved in salt stress, and that AtHSFA7b improves salinity tolerance through inhibiting the expression of these genes.

Some TFs induced by AtHSFA7b positively regulate salt tolerance

AtHSFA7b was observed to differentially regulate 284 transcription factors; among them, 193 TFs were induced by AtHSFA7b, and the others were inhibited by AtHSFA7b. Among the TFs induced by AtHSFA7b, previous studies showed that some of them play roles in salinity tolerance, for instance the transcription factors RAP2.6 (At1G43160), MYB-LIKE 2 (At1G71030), and AtMYB15 (At3G23250) (Ding ; Zhu ; Chan ). However, we hypothesized that other TFs induced by AtHSFA7b would also play roles in salinity tolerance. For confirmation of this hypothesis, we selected seven TFs that were not found to play roles in abiotic stress previously in salt tolerance studies. Among these seven TFs, six played positive roles in salt tolerance, and the other TF is not involved in salt tolerance. Therefore, two conclusions could be drawn here. The first is that bHLH (At5G51780), NAC061 (At3G44350), NAC036 (At2G17040), NAC090 (At5G22380), WRKY38 (At5G22570), and ZFP2 (At5G57520) could positively regulate salt tolerance; however, MYB63 failed to confer salt tolerance to transgenic plants (Fig. 8C–D). The second is that AtHSFA7b improves salt tolerance by inducing a series of TFs, including bHLH, NAC061, NAC036, NAC090, WRKY38, and ZFP2.

Conclusion

Based on the results presented here, a working model of AtHSFA7b responding to salt stress can be proposed. The expression of AtHSFA7b is induced by salt stress, and AtHSFA7b protein then binds to E-box-like motifs and/or HSEs to regulate the genes whose promoter regions contain these motifs, which will directly or indirectly regulate a series of genes to improve salt tolerance. For instance, certain TFs that positively regulate salt tolerance are induced by AtHSFA7b. The NHX family and SOS signaling pathway genes are activated to maintain cellular ion homeostasis; ATHPP2C5 and RHC1 are up-regulated to reduce the stomatal aperture for reducing water loss; GSTs, PODs, and SODs are activated to decrease ROS accumulation; genes such as AtP5CS1 and AtP5CS2 are induced to adjust osmotic potential; the expression of CESA and CSL is also directly or indirectly inhibited by AtHSFA7b, and a reduction in their expression also contributes to improving salt stress tolerance (Fig. 9). These physiological and molecular changes ultimately result in improved salt tolerance.

Fig. 9.

Working model of AtHSFA7b responding to salinity stress. Salt stress induces the expression of AtHSFA7b, and then the activated AtHSFA7b protein interacts with E-box-like and/or HSEs to regulate its target genes, such as SOSs, NHXs, HPP2C5, RHC1, SODs, PODs, GSTs, P5CS, CESAs, CSLs and some transcription factors. These target genes then mediated the following physiological changes: maintaining cellular K+/Na+ homeostasis, improving ROS scavenging capability, enhancing the osmotic potential, and reducing water loss; in addition, cellulose biosynthestic pathways and other unclear pathways mediated by some TFs regulated by AtHSFA7b were also altered, which together leads to improved salt stress tolerance.

Supplementary data

Supplementary data are available at JXB online.

Fig. S1. Phylogenetic study of the HSF proteins from Arabidopsis.

Fig. S2. Analysis of tolerance of AtHSFA7b to thermal stress.

Fig. S3. Hierarchical clustering analysis of the differentially regulated genes.

Fig. S4. Analysis of the ROS scavenging capability conferred by AtHSFA7b.

Fig. S5. The binding of AtHSFA7b to the HSE motif.

Table S1. The primer sequences used for vector construction.

Table S2. The primer sequences used in ChIP-qPCR.

Table S3. The probe sequences used in EMSA.

Table S4. The primer sequences used in qRT-PCR.

Table S5. The genes differentially regulated by AtHSFA7b in RNA-seq.

Table S6. The genes associated with peaks from ChIP-seq of AtHSFA7b.

Click here for additional data file.

Click here for additional data file.

Click here for additional data file.

Click here for additional data file.

Click here for additional data file.

Click here for additional data file.

Click here for additional data file.

Author contributions

YW designed the research; DZ, JW, ZL, and XZ performed the research; DZ, JW, and XZ analysed the data; YW and DZ wrote the paper with contributions from all the authors.

Conflict of interest

The authors declare no conflicts of interest.